{"title":"2026.27 on TeslaDB","link":"https://tesladb.dev/releases/2026.27/","description":"Recent content in 2026.27 on TeslaDB","language":"en-us","last_build_date":"Sat, 19 Sep 2026 00:00:00 +0000","link_json":"https://tesladb.dev/releases/2026.27/index.json","items":[{"title":"FSD (Supervised) v14.3.10","link":"https://tesladb.dev/releases/2026.27/fsd-supervised-v14310/","published_date":"2026-09-19T00:00:00Z","updated_date":"2026-09-19T00:00:00Z","url_feed":"https://tesladb.dev/releases/2026.27/fsd-supervised-v14310/index.json","description":"Full Self-Driving (Supervised) v14.3.10 includes:\nUpgraded the Reinforcement Learning (RL) stage of training the FSD neural network, resulting in improvements in a wide variety of driving scenarios.\nUpgraded the neural network vision encoder, improving understanding in rare and low-visibility scenarios, strengthening 3D geometry understanding, and expanding traffic sign understanding.\nRewrote the AI compiler and runtime from the ground up with MLIR, resulting in 20% faster reaction time and improving model iteration speed.\n"},{"title":"FSD Supervised V14.2 Lite","link":"https://tesladb.dev/releases/2026.27/fsd-supervised-v141-lite/","published_date":"2026-09-19T00:00:00Z","updated_date":"2026-09-19T00:00:00Z","url_feed":"https://tesladb.dev/releases/2026.27/fsd-supervised-v141-lite/index.json","description":"FSD (Supervised) v14.2 Lite includes:\nDistilled the intelligence from HW4 V14 into HW3. This allows HW3 to directly learn how to handle scenarios using HW4 V14 as a guide. This process unlocks the improvements that have been made to HW4 including Reinforcement Learning (RL) and offline models for HW3.\nImproved both proactive and reactive responsiveness across a wide variety of categories including navigation handling, merges and forks, pedestrian interactions, traffic lights, and vehicle cut-in scenarios.\n"},{"title":"Vehicle-to-Home with Powerwall","link":"https://tesladb.dev/releases/2026.27/vehicle-to-home-with-powerwall/","published_date":"2026-09-19T00:00:00Z","updated_date":"2026-09-19T00:00:00Z","url_feed":"https://tesladb.dev/releases/2026.27/vehicle-to-home-with-powerwall/index.json","description":"Your Tesla can now extend your home\u0026rsquo;s backup power duration during an outage by sharing its battery with your Powerwall. When plugged in and enabled, your vehicle automatically supplements your home\u0026rsquo;s energy supply when needed.\nTo enable, go to Charging \u0026gt; Powershare. This feature works when your vehicle is parked and plugged into a compatible Wall Connector with charge above your set discharge limit. Requires a Powerwall.\n"},{"title":"Automatic Collision Evasion","link":"https://tesladb.dev/releases/2026.27/automatic-collision-evasion/","published_date":"2026-09-06T00:00:00Z","updated_date":"2026-09-06T00:00:00Z","url_feed":"https://tesladb.dev/releases/2026.27/automatic-collision-evasion/index.json","description":"Automatic Collision Evasion activates Full Self-Driving (Supervised) to try to keep your vehicle safe and then continue driving. It can engage while you are driving manually and detects that you are not sufficiently attentive to the road (for example, reaching toward the back seat), or that Full Self-Driving (Supervised) may have been unintentionally disengaged.\nSee your Owner\u0026rsquo;s Manual for more details.\n"},{"title":"FSD (Supervised) v14.3.9","link":"https://tesladb.dev/releases/2026.27/fsd-supervised-v1439/","published_date":"2026-09-06T00:00:00Z","updated_date":"2026-09-06T00:00:00Z","url_feed":"https://tesladb.dev/releases/2026.27/fsd-supervised-v1439/index.json","description":"Full Self-Driving (Supervised) v14.3.9 includes:\nUpgraded the Reinforcement Learning (RL) stage of training the FSD neural network, resulting in improvements in a wide variety of driving scenarios.\nUpgraded the neural network vision encoder, improving understanding in rare and low-visibility scenarios, strengthening 3D geometry understanding, and expanding traffic sign understanding.\nRewrote the AI compiler and runtime from the ground up with MLIR, resulting in 20% faster reaction time and improving model iteration speed.\n"}]}